Update sources #893

MattiasBuelens · 2020-08-04T21:57:31Z

This collects the latest updates to the IDL definitions and MDN API descriptions.

Some tooling changes were necessary to deal with a few updated specifications:

Specifications such as Screen Orientation seem to be using a newer version of the W3C spec template, which now includes a <span class="idlHeader">WebIDL</span> element at the top of every WebIDL block (for example in this block). With the current code, we would naively include this header in the IDL content, which broke the WebIDL parser. We now strip this header when fetching new IDLs.
Web IDL has switched from void to undefined, see add a type to represent the undefined value whatwg/webidl#60 and void -> undefined, and allow it anywhere. whatwg/webidl#906. Many specifications have already made the switch, so we need to support this in the emitter.

In order to facilitate review, I've done two baseline update commits: first after running fetch-idl, and then after running fetch-mdn. If you prefer, I can squash these into one commit.

MattiasBuelens · 2020-08-04T22:09:25Z

baselines/dom.generated.d.ts

 interface OES_element_index_uint {
 }

+declare var OES_element_index_uint: {


Hmm, I suppose we would first need #873 to land? These now have [LegacyNoInterfaceObject] in their IDL...

#873 is now merged.

MattiasBuelens

Rebased and reworked! 😁

Note that I cheated a bit with the "Update IDL" commit: I updated all IDLs except for Streams. These need more tweaking and cleanup in our overrides, which I'm doing in #890. 😉

MattiasBuelens · 2020-09-10T20:46:14Z

src/helpers.ts

    ["object", "any"],
    ["sequence", "Array"],
    ["record", "Record"],
+    ["undefined", "void"],


Web IDL has switched from void to undefined, see whatwg/webidl#60 and whatwg/webidl#906. Many specifications have already made the switch, so we need to support this.

For now, I'm mapping this back to TypeScript's void type. However, we could also move this into sameTypes and actually emit undefined in the generated type definitions. Thoughts? 🙂

I like this change except in unions. There is one in CustomElementRegistry:

get(name: string): CustomElementConstructor | void;

You can't use ?. from the value returned by the accessor. This is admittedly a failing of Typescript, but T | void is rare type for all T.

Not sure whether it's easier to fix the emitter or to add an overridingType entry for the accessor

I can add a special case in the emitter to replace void with undefined if it appears in a union type. Right now, it's just for this one method in CustomElementRegistry, but maybe similar methods pop up in the future for which this would apply. 🙂

MattiasBuelens · 2020-09-10T20:51:01Z

inputfiles/overridingTypes.json

        "typedef": [
            {
-                "override-type": "Blob | BufferSource | FormData | URLSearchParams | ReadableStream<Uint8Array> | string",
+                "override-type": "ReadableStream<Uint8Array> | XMLHttpRequestBodyInit",


BodyInit is now defined as:

typedef (Blob or BufferSource or FormData or URLSearchParams or USVString) XMLHttpRequestBodyInit; typedef (ReadableStream or XMLHttpRequestBodyInit) BodyInit;

This change re-aligns our override-type with the new definition.

MattiasBuelens · 2020-09-10T20:53:26Z

inputfiles/idlSources.json

    },
+    {
+        "url": "https://www.w3.org/TR/css-pseudo-4/",
+        "title": "CSS Pseudo-Elements"


This is being used by the Web Animation specification, so we need to include its types. Fortunately, there aren't that many new types.

Neither Chrome nor Firefox provides a stable CSSPseudoElement implementation, so I would just suppress this type.

sandersn · 2020-09-11T16:17:36Z

src/helpers.ts

    ["object", "any"],
    ["sequence", "Array"],
    ["record", "Record"],
+    ["undefined", "void"],


I like this change except in unions. There is one in CustomElementRegistry:

get(name: string): CustomElementConstructor | void;

You can't use ?. from the value returned by the accessor. This is admittedly a failing of Typescript, but T | void is rare type for all T.

Not sure whether it's easier to fix the emitter or to add an overridingType entry for the accessor

sandersn · 2020-09-11T16:19:40Z

inputfiles/mdn/apiDescriptions.json

  "DedicatedWorkerGlobalScope": "The DedicatedWorkerGlobalScope object (the Worker global scope) is accessible through the self keyword. Some additional global functions, namespaces objects, and constructors, not typically associated with the worker global scope, but available on it, are listed in the JavaScript Reference. See also: Functions available to workers.",
  "DelayNode": "The DelayNode interface represents a delay-line; an AudioNode audio-processing module that causes a delay between the arrival of an input data and its propagation to the output.",
-  "DeviceAcceleration": "A DeviceAcceleration object provides information about the amount of acceleration the device is experiencing along all three axes.",
+  "DeviceAcceleration": "REDIRECT  DeviceMotionEventAcceleration",


can this be turned into @deprecated instead of, or in addition to, REDIRECT?

I could change mergeApiDescriptions() to first check if the comment from MDN starts with "REDIRECT", and if so ignore the comment and mark it as deprecated instead.

On another note, is there a reason why Interface does not have a deprecated flag? I kind of expected markAsDeprecated() to also mark the entire interface as deprecated.

sandersn · 2020-09-11T16:36:50Z

baselines/dom.generated.d.ts

+    "webkitanimationend": Event;
+    "webkitanimationiteration": Event;
+    "webkitanimationstart": Event;
+    "webkittransitionend": Event;


this is probably fine, but it's slightly weird to me that webkit* event handlers are being added. Isn't that webkit-specific?

Apparently these have been added to the HTML standard itself, see whatwg/html#5188. 🤷

Should I add them to removedTypes.json? I see there's already a couple of WebKit-specific removals in there.

saschanaz · 2020-09-14T01:29:05Z

Are we confident that this won't introduce any nonexistent or highly experimental APIs? Updating every spec by a single PR makes hard to check them.

MattiasBuelens · 2020-09-14T22:06:50Z

Are we confident that this won't introduce any nonexistent or highly experimental APIs? Updating every spec by a single PR makes hard to check them.

Hmm, good point. I basically ran npm run fetch-idl and npm run fetch-mdn, thinking that this was the usual process. I have not investigated whether all of these new APIs are actually "stable" yet.

I suppose it would be safer to land the tooling changes first, and then create separate PRs for each updated specification, each with their own review?

MattiasBuelens · 2020-09-26T19:16:40Z

Closing this PR in favor of #920, which contains only the tooling changes.

MattiasBuelens requested a review from sandersn as a code owner August 4, 2020 21:57

MattiasBuelens commented Aug 4, 2020

View reviewed changes

MattiasBuelens mentioned this pull request Aug 4, 2020

Improve comments #894

Closed

MattiasBuelens marked this pull request as draft August 6, 2020 22:03

MattiasBuelens added 10 commits September 10, 2020 21:50

Strip header from WebIDL blocks

97c3235

Update IDL

847eea9

Remove redundant entries from removedTypes

aa59123

Fix BodyInit override

1722bb8

Remove removed RTCStatsEvent

bf66528

Support "undefined" in IDL

f51fc8e

Add IDL for CSS Pseudo-Elements

08ad55e

Update baselines

8b15a8f

Update MDN API descriptions

9f5db55

Update baselines

8ee7547

MattiasBuelens commented Sep 10, 2020

View reviewed changes

MattiasBuelens marked this pull request as ready for review September 10, 2020 20:58

MattiasBuelens mentioned this pull request Sep 10, 2020

Update Streams to official WebIDL #890

Merged

sandersn requested changes Sep 11, 2020

View reviewed changes

MattiasBuelens added 3 commits September 12, 2020 01:32

Replace "void" with "undefined" when emitting a union type

dbfd9bf

Deprecate interfaces when their MDN article is a redirect

04b36fb

Remove new WebKit-specific global events

2ce8745

MattiasBuelens mentioned this pull request Sep 26, 2020

Update tooling for new IDL and MDN #920

Merged

MattiasBuelens closed this Sep 26, 2020

Update sources #893

Update sources #893

Uh oh!

Conversation

MattiasBuelens commented Aug 4, 2020 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

MattiasBuelens left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

saschanaz commented Sep 14, 2020

Uh oh!

MattiasBuelens commented Sep 14, 2020

Uh oh!

MattiasBuelens commented Sep 26, 2020

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

MattiasBuelens commented Aug 4, 2020 •

edited

Loading

MattiasBuelens left a comment •

edited

Loading